Page 9 - CE_Industral_Journal_Jan2008

A Novel Approach for Ground Fault Detection

Where r(t) represents the monitored phase and/or Ground Fault Detection using Higher
ground currents. It is assumed that all current record- Order Statistics Based System
ings are corrupted with additive Gaussian noise n(t). Figure 3 is a flowchart showing a higher order statistics
The high impedance ground fault signature is denoted based high impedance ground fault detection system.
by f(t) and represents the instantaneous value of the Acquired data is filtered using a bandpass filter. The
fault current. Normal load signals are denoted by s(t) energy is then calculated and the calculated energy is
and thus Hypothesis H0 represents a non ground fault compared to a threshold to determine if a high imped-
situation and Hypothesis H1 represents a high imped- ance ground fault has occurred.
ance ground fault situation.
The bispectrum and the trispectrum are by definition
the two dimensional and three dimensional Fourier
Data Acquisition transform of the third and fourth order cumulants
defined as,
C 2 (m, n) = E{r (t) r (t + m) r (t + n)} (3)
C 3 (m, n, k) = E{r (t) r (t + m) r (t + n) r (t + k)} (4)
Data Filtering
where E stands for the expected value.
The algorithm implemented in this study is due in part
to Tugnait [11] and utilizes the integrated polyspectra
of single-phase current loads. The detector is developed
such that a detection decision is made using a second
order statistics. This detector uses additional informa-
tion beyond energy signatures.
Energy Calculation
This detector relies on all current spectra including the
in-between harmonics as generated by the pre-process-
ing filter described earlier.
Comparison with
Threshold
Ground Fault Detection using Wavelet
Based System
Figure 4 is a flowchart showing a wavelet based high
Detection Decision impedance ground fault detection system. Acquired
data is filtered using a bandpass filter. Then, as is de-
scribed in detail below, it is decomposed in separate
high and low pass wavelet decomposition filters. The
energy is then calculated and the calculated energy is
compared to a threshold to determine if a high imped-
Figure 3. Higher order statistic based high impedance ance ground fault has occurred.
ground fault detection system.
The following is an application of high impedance
ground fault detection using a wavelet based detection
A detection system and algorithm based on examining system. The continuous wavelet transform of r(t) is
the higher order statistical features of normal currents
has been developed and tested, as discussed below.
Higher order spectra, namely the bispectrum and
trispectrum are traditionally recognized as important
feature extraction mechanisms that are associated with where, the wavelet is (t), is the position and ps is the
the third and fourth order cumulants of random signals. scale.

Industry Journal 8